KR20190136515A - Vehicle recognition device - Google Patents

Abstract

Disclosed is a vehicle recognition apparatus. The vehicle recognition apparatus fixedly installed at a predetermined place to recognize an entering and exiting vehicle comprises: a distance sensor unit measuring a distance between the vehicle recognition apparatus and a vehicle traveling in front of a place where the vehicle recognition apparatus is fixedly installed; an image photographing unit photographing the vehicle; a control unit transmitting a photographing command to the image photographing unit when determined that the vehicle is located at an appropriate location by the distance sensor unit; and a character recognition unit recognizing a license plate of the vehicle from the image photographed by the image photographing unit. Unlike the conventional method, since the character recognition unit does not require a complicated character recognition process in accordance with the time required to calculate the distance from the vehicle and the size of the license plate, so that a complicated algorithm or many operations are not needed in the vehicle recognition apparatus.

Description

Vehicle Recognition Device {VEHICLE RECOGNITION DEVICE}

The present invention relates to a vehicle recognition apparatus, and more particularly, to a vehicle recognition apparatus capable of quickly processing character recognition since a character is recognized in a license plate of a vehicle in a second vehicle after using a distance sensor in a primary lane.

As the modern society demands the efficient management of the complex traffic environment, the necessity of constructing an intelligent traffic management system has increased. In engineering, intelligent traffic management can be divided into intelligent infrastructure system and intelligent vehicle system. Among them, license plate recognition is the most central technology of intelligent infrastructure system, such as highway toll payment system, parking lot access control system, traffic monitoring and urban access control system. It can play a simple and important role to save time by acting on human's simple work in various application fields.

In particular, the recognition technology has been rapidly developed for the efficient management of the complex traffic environment, and is widely used in many places where the camera field is applied, such as illegal parking control and enforcement, signal violation, speed regulation, and parking lot access control.

Among such camera fields, a video control system can be generally applied, and a video control system is intended to utilize a control image captured by a video photographing device such as a video camera installed in a specific space using a closed circuit television (CCTV) for a specific purpose. A system that delivers only to a specific user. At this time, the control image is transmitted using a method of transmitting to the display screen through a closed wired or wireless transmission line.

That is, CCTVs are installed in areas where crime frequently occurs or areas where traffic accidents or fires frequently occur, and the images captured by the camera are transmitted to a central control device for monitoring or storage. This prevents the occurrence of a crime, accident or fire and uses it as a means of resolving a dispute between the parties that may have occurred.

Such CCTV surveillance system monitors the abnormality in the surveillance area by using CCTV. In other words, the surveillance video taken by CCTV is transmitted to the monitor installed in the central management room and displayed. The manager of the central office keeps an eye on the screen displayed on the monitor and monitors for abnormal conditions such as theft or fire.

Mainly in places such as houses, department stores, banks, exhibition halls, etc., the CCTV surveillance system, which is a monitoring device that can prevent theft and easily check the operation status and process flow of the machine, is used.

Such a surveillance camera apparatus is a television system for transmitting an image to a specific receiver, and is connected by wire or wirelessly from a transmission image to a reception image and cannot be arbitrarily received by the general public other than the relevant receiver.

As known, the surveillance camera apparatus checks a surveillance screen displayed on a monitor installed at a specific location by photographing a plurality of surveillance regions to be managed by a plurality of cameras. At this time, if necessary, only a desired scene may be recorded using a video cassette recorder (VCR).

Artificial Intelligence (AI), on the other hand, mimics the human brain and neuronal neural networks, someday allowing computers and robots to think and act like humans. Using this artificial intelligence technology, we are strong in image distortion and lighting conditions such as angle and shape of vehicle license plate image input through camera, and automatic recognition of vehicle number based on reinforcement learning with reliability and recognition rate equivalent to human level. The method is being developed.

In particular, various types of sensors for the license plate recognition of a vehicle are being developed. Among the sensors, a camera as an image capturing sensor may be used. The advantage of the vision sensor using such a camera is that it is easy to detect and recognize the target. However, when using a single camera, it is not easy to calculate the target image that changes according to the distance of the vehicle. there was.

In other words, the recognition of the license plate using artificial intelligence has been steadily studied, but the conventional license plate recognition technologies have a great meaning on the character recognition written on the license plate of the vehicle in the image obtained through the camera. . One of the important factors in the development of a license plate recognition device is how fast it can process information obtained from sensors. In license plate recognition, one problem is to shorten the processing time of video information. In addition, the method of calculating the distance between vehicles based on the character recognition of the license plate has many limitations in view of shortening the image information processing time.

Accordingly, an object of the present invention for solving this problem is a vehicle recognition device fixedly installed at a predetermined place to recognize an entrance vehicle, the vehicle recognition device, and the vehicle of the vehicle traveling in front of the fixed location is installed A distance sensor unit measuring a distance; An image photographing unit photographing the vehicle; If it is determined that the position of the vehicle is located in the proper place by the distance sensor unit, including a control unit for transmitting a shooting command to the image pickup unit and a character recognition unit for recognizing the license plate of the vehicle in the image taken by the image capture unit Unlike the related art, since the character recognition unit does not require a complicated character recognition process according to the time required to calculate the distance to the vehicle and the size of the license plate, it provides a vehicle recognition apparatus that does not need to perform a complicated algorithm or many operations. do.

In order to achieve the above object, the present invention provides a vehicle recognition device fixedly installed at a predetermined place to recognize an entrance and exit vehicle, the distance between the vehicle recognition device and a vehicle traveling in front of the fixed installation place. A distance sensor unit; An image photographing unit photographing the vehicle; If it is determined that the position of the vehicle is located in the proper place by the distance sensor unit, including a control unit for transmitting a shooting command to the image pickup unit and a character recognition unit for recognizing the license plate of the vehicle in the image taken by the image capture unit Provided is a vehicle recognition device.

The vehicle recognition apparatus may include a camera including the distance sensor unit, the image photographing unit, the controller, and the character recognition unit.

The apparatus may further include an image processor converting the analog signal photographed by the image capturing unit into a digital signal and transmitting the digital signal to the character recognition unit.

The distance sensor unit may include a lidar sensor having an FOV angle set to 2 ° to 5 °.

In addition, the character recognition unit, Deep learning license plate detection unit for recognizing the outside of the license plate; A character analyzer after performing binarization on the image of the detected license plate area; It may include a distortion corrector for correcting distortion by detecting warp image transformation and a bound box, and a character classification unit for classifying characters in the bound box into a classifier trained by a convolutional neural network and extracting characters.

According to the vehicle recognition apparatus of the present invention described above, since the character recognition unit does not require a complicated character recognition process according to the time required to calculate the distance from the vehicle and the size of the license plate, a complicated algorithm or many There is no need to perform the operation. That is, the character recognition unit recognizes a license plate of a certain size photographed only when the vehicle enters a desired position, and thus does not have to perform a complicated process for calculating the character recognition according to the size of the license plate. The deep learning recognition process for license plate size recognition can be excluded, which can effectively perform the license plate size recognition.

In addition, since only one license plate recognition camera is installed, there is no need to install a roof sensor or a separate sensor externally through road construction according to the conventional art, and only one license plate recognition camera needs to be installed. There is.

1 is a configuration diagram showing an example of the configuration of a vehicle recognition apparatus according to the present invention.
2 and 3 are views for explaining the function of the distance sensor unit according to the present invention.
4 is a perspective view showing a recognition device according to an embodiment of the present invention, Figure 4 (a) is a perspective view of the recognition device, Figure 4 (b) is an exploded perspective view of the recognition device.
5 is a block diagram and a photograph for explaining the character recognition unit and the event output unit of the present invention.
6 is a flowchart illustrating a sequence according to a vehicle recognition method using a vehicle recognition apparatus according to an embodiment of the present invention.
7 is a flowchart illustrating a method of sequentially performing a character recognition step of the present invention.

In order to fully understand the present invention, preferred embodiments of the present invention will be described with reference to the accompanying drawings. Embodiment of the present invention can be modified in various forms, the scope of the invention should not be construed as limited to the embodiments described in detail below. This embodiment is provided to more completely explain the present invention to those skilled in the art. Therefore, the shape of the elements in the drawings and the like may be exaggerated to emphasize a more clear description. It should be noted that the same members in each drawing are sometimes shown with the same reference numerals. In addition, detailed descriptions of well-known functions and configurations that are determined to unnecessarily obscure the subject matter of the present invention are omitted.

1 is a configuration diagram showing an example of the configuration of a vehicle recognition apparatus according to the present invention.

As shown in FIG. 1, the vehicle recognition apparatus 100 (hereinafter, referred to as a “recognition apparatus”) according to an exemplary embodiment of the present invention may include a distance sensor 110, an image photographing unit 120, an image processing unit 130, and a character recognition unit 140. ) And the controller 160. In addition, the recognition apparatus 100 may further include other components in addition to the above-mentioned components.

For example, the recognition apparatus 100 may be configured by only one camera to perform the long of the present invention, and the above-described components may be included in one camera that is the recognition apparatus 100. In addition, the recognition device 100 may be fixedly installed at a place requiring recognition of a license plate of a vehicle such as an entrance of a parking lot.

In this case, the distance sensor unit 110 measures the position of the vehicle traveling in front of the place where the recognition apparatus 100 is fixed, that is, the distance between the recognition apparatus 100 and the vehicle, and the controller 160 measures the The distance sensor unit 110 determines that the position of the vehicle is in the proper position, and the image capturing unit 120 photographs the vehicle traveling at an appropriate point in front of the place where the recognition device 100 is installed.

That is, according to the present invention, the image photographed by the image capturing unit 120 is not directly transmitted to the image processing unit 130, but first, the control unit 160 is positioned by the distance sensor unit 110 at a proper location. When it is determined that it is, the image capturing unit 120 is characterized in that it has a configuration to obtain an image of the vehicle. Thereafter, the image photographed by the image capturing unit 120 is transmitted to the image processor 130 or the character recognition unit 140 to recognize the license plate of the vehicle. A detailed process thereof will be described later.

2 and 3 are views for explaining the function of the distance sensor unit 120 according to the present invention.

As shown in FIG. 2, the distance sensor unit 120 calculates a distance between the recognition device 100 and the entry / exit vehicle. For example, the distance sensor unit 120 includes a light detection and ranging (LIDAR) sensor, emits radio waves, and measures reflected waves of the emitted radio waves. In this case, the radio wave emitted from the lidar sensor of the distance sensor unit 120 uses a wavelength of 850 nm, and can measure the distance of an object entered within 12 meters within an error range of 6 cm. At this time, the image capturing unit 120 transmits the signal of the radio wave measured by the lidar sensor to the control unit 160.

In addition, the distance sensor unit 110 equipped with a lidar sensor has a field of view (FOV) angle of 2 ° to 5 °, which is relatively small, so that a distance difference of only about 40 cm in the center area of the signal obtained by the rider sensor is provided. Since the sensing can sense only the entrance and exit vehicle of one car entering and exiting the parking lot, and then the image photographing unit 120 performs photographing of the license plate of the vehicle.

The distance sensor unit 110 can be driven at a low power at 1 watt. That is, since the distance sensor unit 110 first measures the distance with the entrance and exit vehicle, there is an effect that can prevent the waste of power consumed to perform a complex operation process in the character recognition unit 140 thereafter.

In addition, as shown in FIG. 3, the distance sensor unit 120 of the recognition apparatus 100 transmits a predetermined signal to the controller 160. For example, when the proper distance between the recognition apparatus 100 and the vehicle is set to 5.7 m, In the detection device 100 fixedly installed at a predetermined place, a signal change occurs at 1400 seconds when the vehicle approaches at a sensing distance of 5.7 m, which is an appropriate distance of the distance sensor 110. The signal is transmitted from the image capturing unit 120 to the control unit 160.

The controller 160 determines the approach of the vehicle based on the signal received from the distance sensor unit 110, and instructs the image processor 130 to acquire the license plate image of the vehicle. Accordingly, unlike the conventional method, since the character recognition unit 140 does not require a complicated character recognition process according to the time required to calculate the distance to the vehicle and the size of the license plate, it is not necessary to perform a complicated algorithm or many operations. It works.

That is, when the distance sensor unit 110 knows in advance the distance between the recognition device 100 and the entering and exiting vehicle, the area of the license plates of the images of the vehicle acquired by the image capturing unit 120 becomes constant to some extent. Since the character recognition unit 140 recognizes a license plate of a predetermined size photographed only when the vehicle enters a desired position, the character recognition unit 140 does not have to perform a complicated process for calculating a character recognition according to the size of the license plate. Deep learning recognition processing for the license plate size recognition required can be excluded, so that the effect can be efficiently performed.

The image capturing unit 120 photographs the access vehicle so that the license plate of the access vehicle is included in the captured image. At this time, the image capturing unit 120 may be composed of only one camera, unlike the prior art. Here, when the image capturing unit 120 is composed of one camera, the resolution per frame of the image obtained by the image capturing unit 120 is constant, and the constant is the area of the license plate constituting the image or This means that the number of pixels in the photographed license plate is constant.

4 is a perspective view showing a recognition device 100 according to an embodiment of the present invention, Figure 4 (a) is a perspective view of the recognition device 100, Figure 4 (b) is an exploded perspective view of the recognition device 100 to be.

As shown in FIG. 4, the recognition apparatus 100 according to an exemplary embodiment of the present invention may be configured with only one camera, and the camera includes a distance sensor unit 110 and an image capturing unit 120. In addition, although not shown, the recognition device 100 may include the controller 160 and the character recognition unit 140 in an embedded form for character recognition using deep learning.

Accordingly, the present invention is a vehicle recognition device in which a character recognition system is embedded and operated together with a rider sensor, which is the distance sensor unit 110, and does not have to be provided with a separate computer, thereby deep learning in the recognition device 100. It has a function to query the vehicle number recognized by itself.

In addition, since the recognition device 100 according to the present invention is provided with only one license plate recognition camera provided with a rider sensor as a distance sensor, a roof sensor or a separate sensor may be installed through a road construction according to the prior art. There is no need to install only one license plate recognition camera, which makes the initial installation simple.

The image processor 130 may convert the analog signal photographed by the image capturing unit 120 into a digital signal and transmit the converted digital signal to the character recognition unit 140. When the image capturing unit 120 is a digital camera, the function of the image processing unit 130 may be included.

5 is a block diagram and a photograph for explaining the character recognition unit 140 and the event output unit 150 of the present invention.

As shown in FIG. 5, the character recognition unit 140 includes a deep learning license plate detection unit 141, a character analysis unit 142, a distortion correction unit 143, and a character classification unit 144. In addition, the character recognition unit 140 may further include an event output unit 150.

The deep learning license plate detection unit 141 processes an image received from the image processor 130 to recognize an outline of the license plate included in the image. In other words, the vehicle license plate area is detected through the regression learning of the convolutional neural network (CNN) and the bound box within the designated region of interest.

In this case, as described above, the image photographed by the image capturing unit 120 is an image photographed by a vehicle of a certain distance, and the size of the license plate is constant, and the number of pixels of the license plate is constant, so that the deep learning license plate detection unit ( The load at 141 can be significantly reduced.

The character analyzer 142 performs binarization on the detected image of the license plate area, performs contour extraction, and then performs a Hough Transform. Thereafter, the distortion correction unit 143 corrects the distortion by detecting a warp image and a bound box. Next, the character classification unit 144 classifies the characters in the bound box into a classifier learned by the convolutional neural network CNN to extract the characters.

Thereafter, the event output unit 150 stores the recognition result of the license plate performed by the character analyzer 142 in real time in the system log of the recognition device 100, and assigns a designated TCP / IP port (for example, 2300 days). The recognition result is transmitted to the control center in real time.

In the above, the configuration and functions of the vehicle recognition apparatus according to the exemplary embodiment of the present invention have been described. Hereinafter, a vehicle recognition method using a vehicle recognition apparatus according to an embodiment of the present invention will be described in detail.

6 is a flowchart illustrating a sequence according to a vehicle recognition method using a vehicle recognition apparatus according to an embodiment of the present invention.

As shown in FIG. 6, a vehicle recognition method using a vehicle recognition apparatus according to an exemplary embodiment of the present invention includes a measurement step S100, a determination step S200, a photographing step S300, and a character recognition step S400.

The measuring step (S100) is a step of measuring a distance between the recognition apparatus 100 fixed to a certain place and an entrance vehicle using a distance sensor unit 110 such as a lidar sensor. For example, in the measuring step S100, the distance sensor unit 110 includes a light detection and ranging (LIDAR) sensor, emits a radio wave, and measures a reflected wave of the emitted radio wave to recognize the device 100. The distance to the entry vehicle can be measured. In this case, the radio wave emitted from the lidar sensor of the distance sensor unit 120 uses a wavelength of 850 nm, and can measure the distance of an object entered within 12 meters within an error range of 6 cm.

In addition, the distance sensor unit 110 equipped with a lidar sensor has a field of view (FOV) angle of 2 ° to 5 °, which is relatively small, so that a distance difference of only about 40 cm in the center area of the signal obtained by the rider sensor is provided. Sensing so that only one vehicle that enters and exits the parking lot can be sensed.

Thereafter, in the determining step S200, the controller 160 determines whether the entrance vehicle enters within a predetermined distance in the measuring step S100, and when determining that the vehicle enters the predetermined distance, the controller 160 takes a picture in the image capturing unit 120. It is a step of transmitting a command.

In the photographing step (S300), the control unit 160 determines that the entrance vehicle has entered within a predetermined distance in the determination step (S200) and the image photographing unit 120 photographs the entrance vehicle when receiving a recording command Step.

In the character recognition step (S400), the character recognition unit 140 recognizes the vehicle license plate based on the image of the vehicle obtained in the photographing step (S300). In this step, unlike the conventional method, since the character recognition step (S400) does not require a complicated character recognition process according to the time required to calculate the distance to the vehicle and the size of the license plate, a complex algorithm or many operations are performed. There is no need to do.

Although not shown, the method may further include converting an analog signal photographed by the image capturing unit 120 to a digital signal in step S300.

7 is a flowchart illustrating a method of sequentially performing the character recognition step S400 of the present invention.

As shown in FIG. 7, the character recognition step S400 includes a license plate detection step S410, a character analysis step S420, a character classification step S430, and an event output step S440.

In the license plate detection step (S410), the image receiving unit 120 in the photographing step (S300) or the conversion step (S350) by processing the image transmitted from the image processing unit 130 in the outside of the license plate included in the image Recognizing this.

The character analysis step (S420) is a step of performing a binarization of the image of the license plate area detected in the license plate detection step (S410), performing contour extraction, and then performing a Hough Transform.

In the character classification step (S430), a character is classified into a classifier learned by a convolutional neural network (CNN) to extract a character in a bound box.

In the event output step S440, the result recognized in the character classification step S430 is stored in the system log of the recognition device 100 in real time, and is recognized as a designated TCP / IP port (for example, 2300). The results are sent to the control center in real time.

The embodiment of the vehicle recognition apparatus of the present invention described above is merely exemplary, and it is well understood that various modifications and equivalent other embodiments are possible to those skilled in the art. Could be. Therefore, it will be understood that the present invention is not limited to the forms mentioned in the above detailed description. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims. It is also to be understood that the present invention includes all modifications, equivalents and substitutes within the spirit and scope of the invention as defined by the appended claims.

100: recognition device 110: distance sensor unit
120: image capture unit 160: control unit
130: image processing unit 140: character recognition unit
141: deep learning license plate detection unit 142: character analysis unit
143: distortion correction 144: character classification unit
150: event output unit S100: measurement step
S200: judgment step S300: shooting step
S400: character recognition step S410: license plate detection step
S410: license plate detection step S420: character analysis step
S430: character classification step

Claims (5)

A vehicle recognition device fixedly installed at a predetermined place to recognize an entrance vehicle,
A distance sensor unit 110 for measuring a distance between the vehicle recognition device and a vehicle traveling in front of a place where the vehicle recognition device is fixedly installed;
An image photographing unit 120 photographing the vehicle;
If the distance sensor unit 110 determines that the location of the vehicle is located in the proper place, the control unit 160 for transmitting a shooting command to the image capturing unit 120 and
Vehicle recognition apparatus comprising a character recognition unit 140 for recognizing the license plate of the vehicle in the image taken by the image photographing unit (120). The method according to claim 1,
The vehicle recognition device,
And a camera including the distance sensor (110), the image capturing unit (120), the control unit (160), and a character recognition unit (140). The method according to claim 1,
And an image processing unit (130) for converting the analog signal photographed by the image capturing unit (120) into a digital signal and transmitting the digital signal to the character recognition unit (140). The method according to claim 1,
The distance sensor unit 110,
A vehicle recognition device comprising a lidar sensor in which the FOV angle is set to 2 ° to 5 °. The method according to claim 1,
The character recognition unit 140,
Deep learning license plate detection unit 141 for recognizing the outside of the license plate;
A character analyzer 142 after binarization of the detected image of the license plate area;
A distortion correction unit 143 for correcting distortion by detecting warp image conversion and a bound box;
And a character classifier 144 for classifying the characters in the bound box into a classifier trained by a convolutional neural network (CNN) to extract the characters.

Images